ruby-changes:26645
From: kosaki <ko1@a...>
Date: Sat, 5 Jan 2013 05:05:52 +0900 (JST)
Subject: [ruby-changes:26645] kosaki:r38696 (trunk): * gc.c (vm_xrealloc): add a few comment why we avoid realloc(ptr,0).
kosaki 2013-01-05 05:05:42 +0900 (Sat, 05 Jan 2013) New Revision: 38696 http://svn.ruby-lang.org/cgi-bin/viewvc.cgi?view=rev&revision=38696 Log: * gc.c (vm_xrealloc): add a few comment why we avoid realloc(ptr,0). Modified files: trunk/ChangeLog trunk/gc.c Index: ChangeLog =================================================================== --- ChangeLog (revision 38695) +++ ChangeLog (revision 38696) @@ -1,3 +1,7 @@ https://github.com/ruby/ruby/blob/trunk/ChangeLog#L1 +Sat Jan 5 05:04:39 2013 KOSAKI Motohiro <kosaki.motohiro@g...> + + * gc.c (vm_xrealloc): add a few comment why we avoid realloc(ptr,0). + Fri Jan 4 20:17:06 2013 Yuki Yugui Sonoda <yugui@y...> * Makefile.in (RBCONFIG): Moved from common.mk in order to use the Index: gc.c =================================================================== --- gc.c (revision 38695) +++ gc.c (revision 38696) @@ -3529,7 +3529,14 @@ vm_xrealloc(rb_objspace_t *objspace, voi https://github.com/ruby/ruby/blob/trunk/gc.c#L3529 if ((ssize_t)size < 0) { negative_size_allocation_error("negative re-allocation size"); } + if (!ptr) return vm_xmalloc(objspace, size); + + /* + * The behavior of realloc(ptr, 0) is implementation defined. + * Therefore we don't use realloc(ptr, 0) for portability reason. + * see http://www.open-std.org/jtc1/sc22/wg14/www/docs/dr_400.htm + */ if (size == 0) { vm_xfree(objspace, ptr); return 0; -- ML: ruby-changes@q... Info: http://www.atdot.net/~ko1/quickml/